arXiv subjects

Danny Bickson

Publications and source records attributed to Danny Bickson.

27 records · Page 2Linked to original sources

arXiv2009

Peer-to-Peer Secure Multi-Party Numerical Computation Facing Malicious Adversaries

We propose an efficient framework for enabling secure multi-party numerical computations in a Peer-to-Peer network. This problem arises in a range of applications such as collaborative filtering, distributed computation of trust and reputation, monitoring and other tasks, where the computing nodes is expected to preserve the privacy of their inputs while performing a joint computation of a certain function. Although there is a rich literature in the field of distributed systems security concerning secure multi-party computation, in practice it is hard to deploy those methods in very large scale Peer-to-Peer networks. In this work, we try to bridge the gap between theoretical algorithms in the security domain, and a practical Peer-to-Peer deployment. We consider two security models. The first is the semi-honest model where peers correctly follow the protocol, but try to reveal private information. We provide three possible schemes for secure multi-party numerical computation for this model and identify a single light-weight scheme which outperforms the others. Using extensive simulation results over real Internet topologies, we demonstrate that our scheme is scalable to very large networks, with up to millions of nodes. The second model we consider is the malicious peers model, where peers can behave arbitrarily, deliberately trying to affect the results of the computation as well as compromising the privacy of other peers. For this model we provide a fourth scheme to defend the execution of the computation against the malicious peers. The proposed scheme has a higher complexity relative to the semi-honest model. Overall, we provide the Peer-to-Peer network designer a set of tools to choose from, based on the desired level of security.

cs.CR↗

arXiv2008

Self-stabilizing Numerical Iterative Computation

Many challenging tasks in sensor networks, including sensor calibration, ranking of nodes, monitoring, event region detection, collaborative filtering, collaborative signal processing, {\em etc.}, can be formulated as a problem of solving a linear system of equations. Several recent works propose different distributed algorithms for solving these problems, usually by using linear iterative numerical methods. In this work, we extend the settings of the above approaches, by adding another dimension to the problem. Specifically, we are interested in {\em self-stabilizing} algorithms, that continuously run and converge to a solution from any initial state. This aspect of the problem is highly important due to the dynamic nature of the network and the frequent changes in the measured environment. In this paper, we link together algorithms from two different domains. On the one hand, we use the rich linear algebra literature of linear iterative methods for solving systems of linear equations, which are naturally distributed with rapid convergence properties. On the other hand, we are interested in self-stabilizing algorithms, where the input to the computation is constantly changing, and we would like the algorithms to converge from any initial state. We propose a simple novel method called \syncAlg as a self-stabilizing variant of the linear iterative methods. We prove that under mild conditions the self-stabilizing algorithm converges to a desired result. We further extend these results to handle the asynchronous case. As a case study, we discuss the sensor calibration problem and provide simulation results to support the applicability of our approach.

cs.DC↗

arXiv2008

A Gaussian Belief Propagation Solver for Large Scale Support Vector Machines

Support vector machines (SVMs) are an extremely successful type of classification and regression algorithms. Building an SVM entails solving a constrained convex quadratic programming problem, which is quadratic in the number of training samples. We introduce an efficient parallel implementation of an support vector regression solver, based on the Gaussian Belief Propagation algorithm (GaBP). In this paper, we demonstrate that methods from the complex system domain could be utilized for performing efficient distributed computation. We compare the proposed algorithm to previously proposed distributed and single-node SVM solvers. Our comparison shows that the proposed algorithm is just as accurate as these solvers, while being significantly faster, especially for large datasets. We demonstrate scalability of the proposed algorithm to up to 1,024 computing nodes and hundreds of thousands of data points using an IBM Blue Gene supercomputer. As far as we know, our work is the largest parallel implementation of belief propagation ever done, demonstrating the applicability of this algorithm for large scale distributed computing systems.

cs.LG↗

arXiv2008

Distributed Kalman Filter via Gaussian Belief Propagation

Recent result shows how to compute distributively and efficiently the linear MMSE for the multiuser detection problem, using the Gaussian BP algorithm. In the current work, we extend this construction, and show that operating this algorithm twice on the matching inputs, has several interesting interpretations. First, we show equivalence to computing one iteration of the Kalman filter. Second, we show that the Kalman filter is a special case of the Gaussian information bottleneck algorithm, when the weight parameter $β= 1$. Third, we discuss the relation to the Affine-scaling interior-point method and show it is a special case of Kalman filter. Besides of the theoretical interest of this linking estimation, compression/clustering and optimization, we allow a single distributed implementation of those algorithms, which is a highly practical and important task in sensor and mobile ad-hoc networks. Application to numerous problem domains includes collaborative signal processing and distributed allocation of resources in a communication network.

cs.IT↗

arXiv2008

Polynomial Linear Programming with Gaussian Belief Propagation

Interior-point methods are state-of-the-art algorithms for solving linear programming (LP) problems with polynomial complexity. Specifically, the Karmarkar algorithm typically solves LP problems in time O(n^{3.5}), where $n$ is the number of unknown variables. Karmarkar's celebrated algorithm is known to be an instance of the log-barrier method using the Newton iteration. The main computational overhead of this method is in inverting the Hessian matrix of the Newton iteration. In this contribution, we propose the application of the Gaussian belief propagation (GaBP) algorithm as part of an efficient and distributed LP solver that exploits the sparse and symmetric structure of the Hessian matrix and avoids the need for direct matrix inversion. This approach shifts the computation from realm of linear algebra to that of probabilistic inference on graphical models, thus applying GaBP as an efficient inference engine. Our construction is general and can be used for any interior-point algorithm which uses the Newton method, including non-linear program solvers.

cs.IT↗

arXiv2008

Gaussian Belief Propagation Solver for Systems of Linear Equations

The canonical problem of solving a system of linear equations arises in numerous contexts in information theory, communication theory, and related fields. In this contribution, we develop a solution based upon Gaussian belief propagation (GaBP) that does not involve direct matrix inversion. The iterative nature of our approach allows for a distributed message-passing implementation of the solution algorithm. We also address some properties of the GaBP solver, including convergence, exactness, its max-product version and relation to classical solution methods. The application example of decorrelation in CDMA is used to demonstrate the faster convergence rate of the proposed solver in comparison to conventional linear-algebraic iterative solution methods.

cs.IT↗

arXiv2008

Gaussian Belief Propagation Based Multiuser Detection

In this work, we present a novel construction for solving the linear multiuser detection problem using the Gaussian Belief Propagation algorithm. Our algorithm yields an efficient, iterative and distributed implementation of the MMSE detector. We compare our algorithm's performance to a recent result and show an improved memory consumption, reduced computation steps and a reduction in the number of sent messages. We prove that recent work by Montanari et al. is an instance of our general algorithm, providing new convergence results for both algorithms.

cs.IT↗

arXiv2008

Peer-to-Peer Secure Multi-Party Numerical Computation

We propose an efficient framework for enabling secure multi-party numerical computations in a Peer-to-Peer network. This problem arises in a range of applications such as collaborative filtering, distributed computation of trust and reputation, monitoring and numerous other tasks, where the computing nodes would like to preserve the privacy of their inputs while performing a joint computation of a certain function. Although there is a rich literature in the field of distributed systems security concerning secure multi-party computation, in practice it is hard to deploy those methods in very large scale Peer-to-Peer networks. In this work, we examine several possible approaches and discuss their feasibility. Among the possible approaches, we identify a single approach which is both scalable and theoretically secure. An additional novel contribution is that we show how to compute the neighborhood based collaborative filtering, a state-of-the-art collaborative filtering algorithm, winner of the Netflix progress prize of the year 2007. Our solution computes this algorithm in a Peer-to-Peer network, using a privacy preserving computation, without loss of accuracy. Using extensive large scale simulations on top of real Internet topologies, we demonstrate the applicability of our approach. As far as we know, we are the first to implement such a large scale secure multi-party simulation of networks of millions of nodes and hundreds of millions of edges.

cs.CR↗

arXiv2008

Gaussian Belief Propagation for Solving Systems of Linear Equations: Theory and Application

The canonical problem of solving a system of linear equations arises in numerous contexts in information theory, communication theory, and related fields. In this contribution, we develop a solution based upon Gaussian belief propagation (GaBP) that does not involve direct matrix inversion. The iterative nature of our approach allows for a distributed message-passing implementation of the solution algorithm. We address the properties of the GaBP solver, including convergence, exactness, computational complexity, message-passing efficiency and its relation to classical solution methods. We use numerical examples and applications, like linear detection, to illustrate these properties through the use of computer simulations. This empirical study demonstrates the attractiveness (e.g., faster convergence rate) of the proposed GaBP solver in comparison to conventional linear-algebraic iterative solution methods.

cs.IT↗